Socioeconomic inequality in postnatal care utilisation among reproductive age women in sub-Saharan African countries with high maternal mortality: a decomposition analysis

Abstract

Objective

To assess the socioeconomic inequality in postnatal care (PNC) utilisation and its contributors among women in 14 sub-Saharan African countries with high maternal mortality.

Design

Community-based cross-sectional study using Demographic Health Survey

Setting

Africa countries with the highest maternal mortality ratio (14 countries)

Participants

All women who had given birth within 2 years prior to the survey (n=64 912)

Primary outcomes

Postnatal care utilisation

Results

The percentage of women who had PNC utilisation was lowest in Ethiopia (23.4%: 95% CI: 22.1%, 24.7%) and highest in The Gambia (91.5%: 95% CI: 90.6%, 92.4%). There was statistically significant pro-rich inequality in the PNC utilisation in all countries except Liberia, meaning PNC utilisation was disproportionately concentrated among women from wealthier households. The weighted Erreygers Normalized Concentration Index (ECI) ranged from 0.0398 in The Gambia to 0.476 in Nigeria; the second-highest inequality was in Cameroon (0.382), followed by Guinea (0.344). The decomposition analysis revealed that the wealth index was the largest contributor to inequality in PNC utilisation in seven countries: Benin, Burundi, The Gambia, Guinea, Nigeria, Sierra Leone, Tanzania. In contrast, educational status emerged as the primary contibutor in Cameroon and Zimbabwe, media exposure in Mali and Mauritania, and distance to healthcare facilities in Ethiopia. However, in Liberia, the weighted ECI of 0.0012 with a p value of 0.96 indicate that there is no significant socioeconomic inequality in PNC utilisation, suggesting that the distribution of PNC utilisation is almost equal across different socioeconomic groups.

Conclusion

Our study revealed a pro-rich inequality in PNC utilisation across all included sub-Saharan African countries with high maternal mortality, except Liberia. This implies that PNC utilisation disproportionately favours the wealthy. Therefore, financially better-off women are more likely to utilise PNC services compared to those who are poor. Addressing the identified contributors of socioeconomic inequalities in PNC utilisation in each country remains crucial for achieving equity in PNC utilisation.

Strengths and limitations of this study.

• Since our study focused exclusively on single-country data analysis, it offers a unique opportunity to examine socioeconomic inequality in PNC utilisation within each of the 14 sub-Saharan African countries. This approach allows us to highlight country-specific health system interventions that have the potential to improve equity in PNC utilisation

• In addition, using the weighted ECI and Wagstaff decomposition analysis, our findings provide precise estimates of inequality in PNC utilisation across these countries.

• However, there are some limitations. As the survey relied on respondents’ reports, there may be a possibility of recall bias.

• Moreover, restricting our study to the most recent birth (last birth) may have underestimated socioeconomic-related inequalities.

Introduction

The postnatal period extends from birth to 42 days afterward,during which the majority of maternal deaths occur.¹ Mothers are at hightened risk during this time due to excessive bleeding, hypertensive disorders, and sepsis.² The World Health Organisation (WHO) report showed that from 2000 to 2020, the global maternal mortality ratio (MMR) declined by 34%, from 342 deaths to 223 deaths per 100,000 live births.³ However, maternal mortality remains unacceptably high. Approximately 287 000 mothers died in 2020, and nearly 95% of all maternal death occurred in sub-Saharan African countries⁴; more than two-thirds of maternal deaths occur in the postnatal period.⁵ The main reasons for high rates of maternal mortality in sub-Saharan African countries are inadequate healthcare facilities, lack of access to healthcare service, and the failure of pregnant women to use available health facilities due to religious beliefs and cultural customs.⁶ However, most of maternal deaths can be prevented with postpartum care services for mothers.⁷

According to the WHO definition, postnatal care (PNC) consists of various preventive services and assessments aimed at identifying and managing complications related to birth during the first six weeks after delivery.⁸ Promoting antenatal care and skilled birth alone is not enough to improve maternal health. Therefore, the WHO recommends that mothers and newborns receive PNC within the first 24 hours after birth, regardless of the place of birth, along with at least three additional visits for both mothers and newborns⁹. Attending all recommended visits helps ensure the health of mothers and newborns by spotting and addressing complications related to childbirth

¹⁰

For several years, strategies to promote universal access to PNC have been recommended, as these interventions can significantly reduce maternal and neonatal mortality.^{11 12} However, despite government programs and policy initiatives, follow-up care after childbirth is remain inadequate. Moreover, many mothers tend to seek PNC only in cases of complications. There are various factors that prevent women from seeking PNC, such as low economic status, lack of education, and poor follow-up services in healthcare.¹³ In addition, previous empirical studies have revealed factors that influence the use of PNC services. These factors include residence, marital status, religion, age, employment status, antenatal care visit, place of delivery and parity.14,20

Several studies examine socioeconomic inequality in maternal health service utilisation in sub-Saharan African countries.21,25 However, only one study looks at PNC as maternal healthcare.²⁶ There is still a lack of studies assessing socio-economic inequality and its contibutors in PNC utilisation. While more than two-thirds of maternal deaths are linked to insufficient utilisation of PNC,⁵ there has not been a systematic analysis of population data to examine the socioeconomic inequality in PNC utilisation in sub-Saharan African countries with high maternal mortality. Such an analysis can help track progress toward Sustainable Development Goals (SDG) 3.1 and 3.2, ensuring that disadvantaged populations are not left behind. Therefore, this study aimed to assess the socioeconomic inequality in PNC utilisation and its contributors in top 14 SSA countries with high maternal mortality using the Demographic Health Survey (DHS).

Methods

Data sources and context

We used data from the DHS collected in 14 sub-Saharan African countries (Benin, Burundi, Cameroon, Ethiopia, The Gambia, Guinea, Liberia, Mali, Mauritania, Nigeria, Sierra Leone, Tanzania, Uganda and Zimbabwe), through a community-based cross-sectional study design. The DHS is a nationwide survey that collects information about maternal and child health, fertility, reproductive health, mortality, nutrition and health behaviours among adults in over 85 countries.²⁷ The countries included in the analysis were chosen due to their high MMR in Africa²⁸ and the availability of recent DHS, as presented in table 1. We used the WHO threshold to classify MMR as follows: <100 (very low), 100–299 (low), 300–499 (high), 500–999 (very high) and over 1000 (extremely high) maternal death per 100,000 live births.³ In our study, we grouped the high, very high and extremely high categories as ‘high’. We then identified the top 14 countries with high MMR using recent DHS data from 2015 onwards to examine socioeconomic inequality in PNC after the Millennium Development Goal agenda were completed in 2015. However, countries with high MMR but without recent DHS data were not included in the analysis.

Table 1. Sub-Saharan African countries with high maternal mortality included in the analysis and sample size.

Country	MMR (per 100 000 live births)	DHS year	DHS sample size
Sierra Leone	1120	2019	3888
Nigeria	917	2018	12 601
Mauritania	766	2021	4502
Liberia	661	2019/20	2233
The Gambia	597	2019/20	3431
Guinea	576	2018	3057
Mali	562	2018	3938
Burundi	548	2016/17	5279
Cameroon	529	2018	3782
Tanzania	524	2022	4360
Zimbabwe	458	2015	2373
Ethiopia	401	2016	4085
Benin	397	2017/18	5428
Uganda	375	2016	5955

DHSDemographic Health Survey

Among the included countries, Sierra Leone had the highest MMR at 1120 deaths per 100,000 live births, while Uganda had the lowest at 375 deaths per 100,000. Countries with the highest MMR often lack good healthcare infrastructure, including a shortage of skilled provider, insufficient medical facilities, and limited access to prenatal and PNC. They also face issues like high fertility rate, low life expectancy, political instability, and humanitarian crises. Overall, these countries have weak health systems with limited funding, poor governance, and inadequate healthcare financing.^{29 30}

Sampling procedures and sample size

The DHS programme employs a two-stage sampling design. The countries are divided into clusters or enumeration areas, which serve as the primary sampling units (PSUs). The selection of PSUs is typically based on a sampling frame derived from recent census data or other population information sources. A subset of clusters is selected using probability proportional to size sampling. The size of each cluster is proportionate to its population size. The number of selected clusters may vary depending on the desired sample size and the survey objectives. Within each selected cluster, a systematic sampling method is used to select households. This involves creating a list of households within each cluster and systematically selecting a predetermined number of households from the list. Within each selected household, eligible participants are identified based on the survey criteria. For this study, we used individual record data sets containing information from all eligible women aged 15 to 49 years. The source population for this study included all women who had given birth within 2 years prior to the survey. In total, 64 912 mother were part of this analysis, with a sample sizes ranging from 2233 (Liberia) and 12 601 (Nigeria) (see table 1). Data from each country were analysed separately.

Measurement of variables

PNC utilisation is defined as women who reported receiving PNC either at a health facility or at home from a skilled provider (doctor, nurse, midwife, health officer, auxiliary nurse and community nurse) within the first 48 hours after giving birth.³¹ The PNC utilisation was binary and was coded as ‘1’ if women used PNC and ‘0’ otherwise.

Our study includes individual and community-level variables.^{16 18 20 32} Individual-level variables include the age of the mother recorded in completed years (15–24, 25–34, 35+), employment status (not working, professional, agriculture and others), marital status (married, not married), head of household (male, female), educational status of mother (no education, primary, secondary and higher), educational status of husband (no education, primary, secondary and higher) and parity (1–5 and 5+). Wealth status (poorest, poorer, middle, richer, richest) is included as a household-level variable.

In addition, the DHS assessed woman’s autonomy in healthcare decisions by asking ‘who usually decides on your healthcare?’ The responses were coded as follows: ‘1’ for when the women decided alone; ‘2’ for joint decisions with their partner, and ‘3’ for when the partner decided alone.³³ Women who made decisions alone or with their partner were recorded as ‘1’, while those whose partners made decisions alone were recoded as ‘0’. In this contex, ‘0’ indicates a lack autonomy in healthcare decision-making, while ‘1’ indicates autonomy.

The variable ‘pregnancy wantedness’ was divided into three categories: ‘wanted then’, ‘wanted later’ and ‘wanted no more’. ‘Wanted then’ indicates that women desired the pregnancy at the time it happened. ‘Wanted later’ means women did not initially want the pregnancy, but accepted it over time. ‘Wanted no more’ refers that the pregnancy was entirely unwanted.³⁴ The other variable is ‘desire to have more children’. This was assessed in DHS by asking currently married women ‘Do you want to have a/another child?’ The possible responses were (1) wants within 2 years, (2) wants after 2 years, (3) undecided, (4) want no more and (5) sterilised respondent or partner). We recoded the response so that women who answered as undecided, want no more, or sterilised (respondent or partner) were given a code of ‘0’, while those who answered ‘wants within 2 years’ or ‘wants after 2 years’ recieved a code of ‘1’. Here, ‘0’ indicates no desire for more children, and ‘1’ indicates a desire to have more children. Moreover, three variables were used to assess the mother’s media exposure status: listening radio, reading a newspaper and watching television, and labelled as ‘yes’ if mother has exposure to either of the three media sources at least once a week or ‘no’ if a mother has exposure to none of them.²⁷

While, the community-level variables include a place of residence (rural, urban), and the difficulty of getting nearest health services (big problem, not big problem). The mothers were asked whether the distance to the health facility a significant problem for them was or not, to assess the difficulty of reaching the nearest health facility.

Equity stratifier/dimension of inequality

PNC utilisation inequality was measured using economic status as an equity stratifier. Economic status was assessed using data from DHS through a wealth index, which considers various assets owned by households. Examples of these assets include farmland, livestock, building materials, radios/TVs, refrigerators, and sanitation infrastructuree. The wealth index was created using principal component analysis for urban and rural areas separately. This approach generates uncorrelated components, with the first principal component explaining the most variability in the data.³⁵ The asset score, a continuous variable, rank households from the lowest to the highest and wealth index is divided into five categories: poorest, poorer, middle, richer and richest.³⁶

Data management and statistical analysis

The data from 14 countries were extracted, cleaned, recoded and analysed separately using STATA V.16 statistical software. We presented descriptive statistics through tables, figures and narratives. We conducted a weighted data analysis using individual sample weights for women to consider the complex design of the DHS multi-stage cluster sampling. In addition, we adjusted our analyses for clustering and stratification using the svyset Stata command.

To assess the socioeconomic inequality in PNC utilisation, a concentration index (CI) was calculated. The index for an unbound variable varies from −1 to 1; for bounded variables, it ranges from µ−1 to 1−µ.³⁷ Healthcare inequality is decomposed based on the assumption that the health variable is a linear function of the explanatory variables. In this study, the health variable is PNC utilisation, which is a binary variable that ranges from 0 to 1. Therefore, we used Erreygers Normalized Concentration Index (ECI), that, a modified version of the CI.³⁸ Mathematically, ECI can be defined as:

$${\displaystyle \mathrm{E}\mathrm{C}\mathrm{I}=4\times \mu \times \mathrm{C}\mathrm{I}(\mathrm{y})}$$

where ECI is Erreygers concentration index, µ is the mean of the PNC utilisation and CI(y) is the generalised concentration index. Then, the ECI with the SE was reported in this study.

The concentration curve was used to show socioeconomic inequality in PNC utilisation graphically. The curve demonstrates the cumulative share of PNC utilisation on the y-axis against and the cumulative share of women ranked by the wealth index on the x-axis, arranged from the poorest to richest. If the concentration curve lies at a 45° line (the line of perfect equality), everyone will have the same condition for PNC utilisation regardless of wealth status, indicating that there is no socioeconomic inequality (ECI=zero). On the other hand, when the curve lies above the line of equality (when the ECI takes a negative value), the PNC utilisation will be concentrated among the poor (pro-poor). However, if the ECI value is positive, the curve will be below the line of equality, indicating the PNC utilisation is concentrated among the rich (pro-rich).³⁹ Therefore, a concentration curve can be examined visually to determine if it is above or below the line of equality. The ECI with its p value was calculated to assess the statistical significance of the difference between the concentration curve and the line of perfect equality.

A modified Wagstaff decomposition method was used to identify the relative contribution of different factors to the socioeconomic-related inequality in PNC utilisation.^{37 39} In any linear additive regression model predicting a health outcome (y)³⁹:

$${\displaystyle {\displaystyle \mathrm{y}=\mu +\sum _k{\beta }_k{X}_k+\in }}$$

The concentration index for y, CI, is given as:

$${\displaystyle {\displaystyle y=\sum _k\left(\frac{{\beta }_k{\stackrel{-}{\mathrm{X}}}_k}{\mu }\right)C_K+\frac{g{c}_{ϵ}}{\mu }}}$$

where ‘y’ is the socioeconomic inequality of PNC utilisation, $X_k$ is a set of the socioeconomic determinants of PNC utilisation, ${\beta }_k$ is the coefficient of $X_k$, $\mu$ is the mean of y, ${\stackrel{-}{\mathrm{X}}}_k$ is the mean of $X_k$, $C_K$ is the CI for $X_k$, ${gc}_{ϵ}$ is the generalised CI for the error term ($\in$), ${\displaystyle \frac{{\beta }_k{\stackrel{-}{\mathrm{X}}}_k}{\mu }}$ is the elasticity of y with respect to ${\stackrel{-}{\mathrm{X}}}_k$.⁴⁰

Patient and public involvement

We used publicly available secondary data for the analysis; patients and the public were not directly involved in the design, analysis and interpretation of the findings.

Results

Sociodemographic and economic characteristics of respondents

The mean age of respondents was highest in Burundi (29.1 years) and lowest in Cameroon and Zimbabwe (26.9 years). The highest proportions of women were in the age group 24–34 years in all countries included in the analysis except Liberia (online supplemental table 1). Urban residents accounted for the majority of respondents in Sierra Leone (69.44%), while rural residents dominated in the remaining 13 countries, with the highest numbers in Burundi (83.6%), followed by Uganda (81.7%) and Ethiopia (79.4%). The proportion of participants who did not attend formal education ranged from 0.97% in Zimbabwe to 74.9% in Guinea.

Postnatal care utilisation

The PNC utilisation was lowest in Ethiopia (23.4%: 95% CI: 22.1%, 24.7%) and highest in The Gambia (91.5%: 95% CI: 90.6%, 92.4%). The percentage of PNC utilisation exceeded 50% in 12 of the 14 countries. In Ethiopia (23.4%) and Nigeria (45.4%), fewer than half of the women used PNC (figure 1). As shown in figure 1, the level of PNC utilisation was highest among women in the richest household quintile across all countries.

Figure 1. Proportion of women using postnatal care, by wealth index, by country.

Socioeconomic-related inequality in postnatal care utilisation

The results of the ECI analyses are displayed in online supplemental table 2. There was statistically significant pro-rich inequality in the PNC utilisation in all countries except Liberia; meaning PNC utilisation was disproportionately concentrated among women from richer households. The weighted ECI ranged from 0.0398 in The Gambia to 0.476 in Nigeria; the second-highest inequality was in Cameroon (0.382), followed by Guinea (0.344) and Mauritania (0.330). Concentration indices were less than 0.1 in three countries (Sierra Leone, The Gambia and Liberia), indicating a relatively equitable utilisation of PNC. However, in Liberia, the weighted ECI of 0.0012 with a p value of 0.96 indicate that there is no significant socioeconomic inequality in PNC utilisation; suggesting that the distribution of PNC utilisation is almost equal across different socioeconomic groups. The concentration curves for PNC utilisation for all countries are displayed in figure 2.

Figure 2. Concentration curves for postnatal care utilisation in 14 sub-Saharan African countries with high maternal mortality.

Decomposition of socioeconomic-related inequality in PNC utilisation

The decomposition analysis identifies the contribution of each determinant to the overall socioeconomic inequality in PNC utilisation. Key metrics such as coefficient (along with thier significant levels), elasticity, CI and percent contribution were calculated to determine the factors that contribute to socioeconomic inequality (online supplemental table 3).

The decomposition analysis revealed that the wealth index was the largest contributor to inequality in PNC utilisation in seven countries: Benin, Burundi, The Gambia, Guinea, Nigeria, Sierra Leone, Tanzania. In contrast, educational status emerged as the primary contibutor in Cameroon and Zimbabwe, media exposure in Mali and Mauritania, and distance to healthcare facilities in Ethiopia. The second largest contributor to inequality in PNC utilisation varied across countries. Media exposure was the second most significant factor in Benin, Cameroon, The Gambia and Tanzania, while the wealth index held this position in Ethiopia, Zimbabwe, Uganda and Mauritania. In Guinea and Mali, place of residence was the second largest contributor, and in Tanzania, it was parity. Notably, in Burundi, Nigeria and Sierra Leone, educational status ranked as the second largest contributor to inequality in PNC utilisation.

The elasticity was assessed to measure the change in PNC utilisation with a one-unit change in the independent variables.^{41 42} The elasticity value could be positive or negative, signifying an increase or decrease in PNC utilisation corresponding to a positive change in the independent variables. Importantly, the distance to health facilities showed negative elasticity in all countries except Mauritania and Nigeria, suggesting that women who faced significant challenges in accessing health facilities were less likely to utilise PNC. Similarly, parity during the index pregnancy exhibited negative elasticity across all countries, except for Burundi, Cameroon, and Uganda, indicating that the higher parity was associated with reduced likelihood of PNC utilisation. In contrast, educational status demonstrated positive elasticity in all countries, except The Gambia, Mauritania, and Uganda, implying that women with formal education were more likely to have PNC utilisation.

Discussion

This study highlights a significant pro-rich inequality in PNC utilisation, with varying level of inequality across the 14 countries. These findings align with previous studies conducted in Cambodia,⁴³ Ghana,⁴⁴ Western China,^{45 46} Zimbabwe,⁴⁷ Namibia,⁴⁸ Sri Lanka,⁴⁹ India,¹³ Bangladesh⁵⁰ and Ethiopia.^{51 52} Evidence indicates a strong association between PNC utilisation and economic class; better PNC utilisation being more prominent among the rich than the poor.⁵⁰ This disparity suggests that economically disadvantaged women face significant barriers to accessing PNC services, hindering their ability to benefit from universal PNC access. To address this issue, it is essential to promote intersectoral collaboration among development sectors and implement system-wide integrated activities aimed at reducing poverty, enhancing PNC service, and promoting equity.

Moreover, the decomposition analysis showed that the inequalities in PNC utilisation were mainly attributed to women’s wealth status in seven countries followed by educational status in five countries and media exposure in four countries. Previous studies have also documented that wealth is a significant factor in PNC utilisation, suggesting that being in a rich household substantially increases the uptake of PNC services.2053,55 Women who cannot afford to pay for direct and indirect medical costs still face difficulties to access health facilities, even though PNC services are provided for free in most areas.⁵⁶ The primary concern of poor households is meeting their basic needs, leaving them with limited financial resources for healthcare expenses.⁵⁷ Therefore, it is crucial for the government to implement interventions and initiatives as well as to expand existing community-based programmes aimed at empowering women to generate their own income. This would enhance their economic self-sufficiency, and improve their access to health services.

This study revealed that women educational status is a significant contributor to the overall socioeconomic inequality in PNC utilisation. Previous studies have also highlighted a strong positive relationships between educational attainment and the utilisation of PNC.45,4749 51 58 59 The possible explanation could be that higher educational attainment often correlates with better economic resources, enabling women to take control of their health and facilitating easier access to health services. Education enhance individuals’ understanding of the benefits of preventive healthcare and increase awareness of available health services. Moreover, it improves individuals’ capacity to promote health by influencing their lifestyle and enhancing the utilisation of healthcare services through better knowledge, attitudes and practices.⁶⁰ This implies that program planners and decision-makers should consider targeted interventions to minimise socioeconomic inequality in PNC utilisation. This could involve increasing access to education to enhance women’s educational status.

Our study showed that media exposure is a positive contributor to socioeconomic inequality in PNC utilisation. This finding is consistent with the studies done in Ethiopia,^{52 58} Benin,⁵⁵ Bangladesh⁶¹ and South Asia.⁶² This may be because mass media serve as an important means of disseminating health information, which can increase women’s knowledge and attitudes toward health service utilisation.⁶³ As a result, this can lead to an improved service utilisation, including PNC. Moreover, mass media is a popular and cost-effective public health promotional tool that influences women’s health care-seeking behaviour by increasing health knowledge and attitudes toward healthy living.⁶⁴

Strengths and limitations of the study

Since our study focused exclusively on single-country data analysis, it offers a unique opportunity to examine socioeconomic inequality in PNC utilisation within each of the 14 sub-Saharan African countries. This approach allows us to highlight country-specific health system interventions that have the potential to improve equity in PNC utilisation. In addition, by using ECI, tailored for binary outcomes, our findings provide precise estimates of inequality in PNC utilisation across these countries.

However, there are some limitations. Our source population includes all women who have given birth within the 2 years prior to the survey. This timeframe may pose challenges for women when recalling specific details related to the questions asked by the data collectors, leading to potential recall bias. In addition, restricting our study to the most recent birth (last birth) may have underestimated socioeconomic-related inequalities.

Conclusions

Our study revealed a pro-rich inequality in PNC utilisation across all included sub-Saharan African countries with high maternal mortality, except Liberia. This implies that PNC utilisation disproportionately favours the wealthy. Therefore, financially better-off women are more likely to utilise PNC services compared to their poorer counterparts. It is essential to address the identified contributors of socioeconomic inequalities in PNC utilisation in each country achieve equity in access to these service.

Data availability statement

All data relevant to the study are included in the article or uploaded as online supplemental information.